3D Stereo Imaging Technology From 2D Image Sequence

Abstract

In this paper, we will analyze the depth factor in general monocular video sequence. We can distinguish foreground from background without additional information and then create the binocular image by shifting foreground pixels. By applying CID method, we will get the strength of the sharpness and contrast from image by evaluating the farness of the region based on property of region color, and then a depth map of the image can be generated. An optical flow method is also applied to search and compare the block motion status between previous and current frame. The motion vector of block can be estimated by calculating the similarity of blocks based on least square error, and then the farness of block can be estimated by the amount of block motion. Finally, the static and motion depth information are integrated to estimate more accurate depth map. By shifting foreground pixel by depth amount, a binocular image pair can be created. We will perceive 3D stereo feeling without putting glasses on by means of a parallax barrier stereoscopic display.